We propose a novel geometric and photometric 3D mapping pipeline for accurate
and real-time scene reconstruction from monocular images. To achieve this, we
leverage recent advances in dense monocular SLAM and real-time hierarchical
volumetric neural radiance fields. Our insight is that dense monocular SLAM
provides the right information to fit a neural radiance field of the scene in
real-time, by providing accurate pose estimates and depth-maps with associated
uncertainty. With our proposed uncertainty-based depth loss, we achieve not
only good photometric accuracy, but also great geometric accuracy. In fact, our
proposed pipeline achieves better geometric and photometric accuracy than
competing approaches (up to 179% better PSNR and 86% better L1 depth), while
working in real-time and using only monocular images.

提出了一种新的基于几何和光学的三维地图制图方法，利用稠密单目 SLAM 和实时分层体积神经辐射场技术，通过提供估计准确的位姿和深度图的相关不确定性，实时适应神经辐射场的场景。通过采用提出的基于不确定性的深度损失，实现了良好光学和几何精度，在实时和单目成像的情况下比竞争方法更精确 (最高提高 179% 的 PSNR 和 86% 的 L1 深度)。

NeRF-SLAM：神经辐射场实时稠密单目 SLAM

NeRF-SLAM: Real-Time Dense Monocular SLAM with Neural Radiance Fields

We present the second public data release of the Dark Energy Survey, DES DR2,
based on optical/near-infrared imaging by the Dark Energy Camera mounted on the
4-m Blanco telescope at Cerro Tololo Inter-American Observatory in Chile. DES
DR2 consists of reduced single-epoch and coadded images, a source catalog
derived from coadded images, and associated data products assembled from 6
years of DES science operations. This release includes data from the DES
wide-area survey covering ~5000 deg2 of the southern Galactic cap in five broad
photometric bands, grizY. DES DR2 has a median delivered point-spread function
full-width at half maximum of g= 1.11, r= 0.95, i= 0.88, z= 0.83, and Y= 0.90
arcsec photometric uniformity with a standard deviation of < 3 mmag with
respect to Gaia DR2 G-band, a photometric accuracy of ~10 mmag, and a median
internal astrometric precision of ~27 mas. The median coadded catalog depth for
a 1.95 arcsec diameter aperture at S/N= 10 is g= 24.7, r= 24.4, i= 23.8, z=
23.1 and Y= 21.7 mag. DES DR2 includes ~691 million distinct astronomical
objects detected in 10,169 coadded image tiles of size 0.534 deg2 produced from
76,217 single-epoch images. After a basic quality selection, benchmark galaxy
and stellar samples contain 543 million and 145 million objects, respectively.
These data are accessible through several interfaces, including interactive
image visualization tools, web-based query clients, image cutout servers and
Jupyter notebooks. DES DR2 constitutes the largest photometric data set to date
at the achieved depth and photometric precision.

Dark Energy Survey 第二次公共数据发布包括光学 / 近红外成像数据、基于单次观测和联合图像的源目录、以及从 6 年的 DESSCOP 科学操作中汇编的相关数据产品。该数据集包含来自 DES 广角度差异覆盖的约 5000 平方度的数据，结果显示：最高精度为约 10mmag，包含约 691 million 的天体目标。